Hello, I’m working with the full count 1870 census and noticed some disparities between the IPUMS data and the federal census as available on census.gov or ancestry.com. Namely, I was trying to find the number of Chinese people in various states in 1870, according to their listed race and birthplace. I noticed for Mississippi, when I had my code count for Chinese (RACE == 4), the count was 0, whereas on the census the count is 16. I then tried to count for birthplaces listed as China with the variable BPL == 500, and got back 318, which is way higher than anything I have seen on the census indicate. I was wondering if perhaps there’s something to explain this discrepancy or if there is a way I could see where IPUMS got this data from so I could check in census scans.
Thank you
In general, we don’t expect IPUMS to always match census estimates exactly, even with the full count data. This is due to uncertainty around processing that the Census Bureau may have done to their historical estimates as well as transcription error, which may intersect with processing by the Census Bureau.
In the 1870 IPUMS full count data, I confirmed that there are no persons coded as Chinese in the RACE variable in Mississippi (using STATEICP==46). I then tabbed RACE for all cases born in China (BPL = 500) and see 17 persons, all of whom have Black as their race; I am unsure of how you arrived at a count of 318, but I was not able to replicate this. This count of 17 persons with a birthplace in China closely matches what you found in data direct from Census which has 16 persons coded as Chinese.
As a note, 1870 is the first year that Chinese was recorded in the Census and was originally categorized as “color.” Other race/color categories enumerated during the 1870 census were “White (W), black (B), mulatto (M), Chinese (C), Indian (I).” You can find more details on the enumeration form, the census questions and enumerator instructions at IPUMS.
